Teens and young adults who are prescribed high doses of antidepressants are twice as likely as those prescribed more moderate doses to commit suicide, according to a study published in JAMA Internal Medicine this week.
A 2006 US Food and Drug Administration (FDA) meta-analysis of placebo-controlled antidepressant clinical trials found an elevated risk of suicide in teens and young adults, raising serious concerns about the safety of the medications for patients in this age group. However, many questioned the validity of the finding and suggested that the underlying depression itself may have been responsible for the elevated suicide risks—setting off years of debate about the potential risks of antidepressants for young patients.
Critics of the FDA findings have argued that patients and physicians may be fearful of using these medications based on unclear risks, inadvertently exposing patients to possibly more substantial risks related to untreated depression.
The new study, an analysis of medical records from more than 150 000 US individuals who took antidepressants between 1998 and 2010, appears to bolster the link between antidepressants and youth suicide by demonstrating a dose-dependent response. The researchers found that the risk of suicide among individuals 24 years or younger who took higher-than-average doses of the medications was twice that of patients who took average doses. This would translate into 1 additional suicide for every 136 patients taking a higher-than-average dose. There was no difference in the suicide risk between high-dose and moderate-dose antidepressant use in individuals older than 24 years.
Matthew Miller, MD, ScD, associate professor of health policy and management at the Harvard School of Public Health, discussed the findings with news@JAMA.
news@JAMA: Why did you decide to do the study?
Dr Miller: We were motivated by the FDA meta-analysis of placebo-controlled trial data that suggested antidepressants may increase the risk of adolescent and young adult suicide. It led us to ask whether higher doses might lead to the same effect. We know higher doses aren’t better than [moderate] doses, so would we see a suicide effect related to age at higher doses.
news@JAMA: Could the increased risk you observed in the higher-dose group explain the elevated suicide risk seen in the FDA analysis?
Dr Miller:We don’t address that question. We asked is there a dose response? We have shown the dose that children and young adults receive matters. It’s consistent with what the FDA observed in the placebo-controlled trials. In the absence of dose-dependent response, you would have questioned the FDA’s finding. The existence of a dose-response in youth suicide risk in this analysis is a sobering finding. It’s consistent with the FDA’s black box warning, and clinicians should be alert to potential antidepressant dose–related suicide risks in adolescents and young adults.
news@JAMA: What do you think are the clinical implications of your findings?
Dr Miller:One should start children and young adults at lower or moderate doses, if it is decided to use antidepressants. Young patients taking antidepressants should be monitored closely for months. This study should start discussions between physicians and patients about alternatives to these antidepressants.
news@JAMA: With observational studies there is always concern about potential confounders. How did you address this?
Dr Miller: One thingwe did was take extraordinary pains to make apple-to-apple comparisons. Some news reports question whether the 2 groups were different enough to explain the findings. We found a 2-fold difference in suicide risk between the 2 groups, and we controlled for lots of potential confounders. We did a propensity analysis that would identify any known suicide risk as a confounder. We did a bias analysis to determine how strong an unknown confounder would have to be and how unbalanced our groups would have to be for an unknown confounder to explain the result. Based on this analysis, an unmeasured confounder would have to be a stronger predictor of suicide than any known risk factor. We feel it would be unlikely that such a risk factor exists.
news@JAMA: What additional research is needed?
Dr Miller:One study that is needed would look at whether increasing the dose of an antidepressant for patients who are not doing well on a usual dose helps or hurt in terms of deliberate self-harm. Our study doesn’t look at dose escalation vs changing antidepressants or switching to another therapy. Those would be other important questions to ask.